The present invention relates to distributed data processing, and more specifically, this invention relates to prioritizing recovery-related input/output (I/O) operations for distributed computing jobs.
Distributed computing environments perform valuable data processing on a regular basis. Many different I/O operations are requested and performed within distributed computing environments during their regular operation. However, current I/O operation scheduling fails to account for an importance of particular I/O operations such as checkpointing I/O operations and recovery-related I/O operations when compared to normal workload I/O operations within the distributed computing environment.